In college I majored in something called 'Social Studies,' which was a catch-all term for a hybrid social sciences degree. I studied a bit of everything: economics, statistics, intellectual history, sociology, philosophy. But I ended up taking the most courses in anthropology.
Anthropology is the study of how humans derive meaning from the world around them. I joke that one of the reasons I ended up in venture was because I loved anthropology in college. Technology is a rich medium for understanding how humans ascribe meaning and behave as a result.
Just take Instagram as an example. A like, an emoji react, or a DM all carry their own kind of social significance and allow us to gesture to one another in discrete and nuanced ways. "Pics or it didn't happen," "may delete later," and "soft launch" are all examples of new behaviors that technology has altered and evolved from time-worn human instincts.
As humans we make meaning of many things. And anthropology taught me that one of things humans have always found endlessly meaningful is the study of one's self. This is why I get excited about technologies today that have the ability to teach us more about who we are.
Of course this is not a new idea. For decades platforms have been teaching us about ourselves by digesting our user data and regurgitating it back to us in the form of personalized insights and recommendations (and ads). Youtube has a sense of what I like to watch, Amazon what I like to buy, and Tiktok what I find funny. But innovations like Spotify Wrapped (a yearly highlight for me) are signs that more can be done with data to woo us users with our own reflections.
Things get interesting when considering this latest tranche of AI-native businesses, which are jostling left and right to compete for our user data to improve their LLMs and build defensibility.
Traditionally, startups have had three main avenues to access our data:
1/ They monitor us on their platforms and gather data about us slowly and gradually over time (e.g., TikTok monitoring what content we share with friends)
2/ They ask us for data like our emails, calendar, or photo roll (e.g., Superhuman asking us for access to our gmail)
3/ They invite us to tell them about ourselves, generating a tailored data set that they ideally use to serve us better (e.g., Goodreads asking us what genres we like to read)
The problem with #1 is that it's slow and already requires a user to be using the app. The problem with #2 is that it can feel invasive and requires user trust. The problem to date with #3 has been that it can often feel like a chore for users.
But there's creativity happening in #3. Platforms are getting more imaginative about designing user experiences that solicit data from users in return for insights that actually feel valuable.
For example, there's a dating app called Snack that uses AI to set users up on dates. Users train a personal agent on who they are as an individual and what they want out of a romantic partner. This match-making agent then goes out into the world and comes back with a set of potential matches.
How does it know what users want? It asks them. And not through another dreary questionnaire. Instead Snack offers users a personality test. In exchange for answering two-dozen questions about themselves users receive a personalized summary of what kind of 'snack' they are, covering their romantic identity, attachment style, and ideal partner types.
This kind of curated experience not only builds good will, thus encouraging users to provide even more data about themselves. But also, by asking for a small amount of 'work' upfront, it makes users feel like participants in their own well-being. Some say the Netflix 'surprise me' button failed because it was almost too frictionless a feature. From the perspective of behavioral psychology, we as humans are more likely to value an asset that we feel like we have earned, as opposed to one that's landed in our lap. A little bit of friction can go a long way.
Anthropology taught me that humans are not only fascinating, but fascinated by themselves. The most successful AI applications won't just ask humans to hand over their data, they'll court them for it. And one of the best ways to do that is to reorganize that data into a pixelated mirror of who that human is, maybe even for free. Did you know you really liked this song? Did you know this color really suits you? Did you know you feel calmer after talking to this person? Companies that do this, and do so transparently and collaboratively, will come away with a lot of good will. And the data to stoke it.
